The problem here is that you have a different PATH depending on whether
you login interactively or non-interactively. Specifically:
ssh node "echo $PATH"
gives a different result than
ssh node
node$ echo $PATH
Check out the LAM FAQ under "Booting LAM" -- there's info in there
about setting up your shell startup files properly (e.g., to ensure
that you interactive and non-interactive paths are either the same or
both include the bin directory where LAM/MPI is installed).
Hope that helps.
On Jul 13, 2005, at 5:02 PM, Tony Arcieri wrote:
> We've been using lam 7.1.2b18 successfully for several months as 7.1.1
> has
> formatting issues on OS X. Recently we reinstalled MacOS X 10.3.9
> Server
> on the head node of our cluster, and upon reinstalling LAM MPI
> 7.1.2b18 and
> trying to lamboot we get the following:
>
> ERROR: LAM/MPI unexpectedly received the following on stderr:
> bash: line 1: hboot: command not found
> -----------------------------------------------------------------------
> ------
> LAM failed to execute a LAM binary on the remote node "node2".
> Since LAM was already able to determine your remote shell as "hboot",
> it is probable that this is not an authentication problem.
>
> Absolutely nothing had changed on node2 since it was last working, only
> node1 has been reinstalled...
>
> Now here's the real kicker...
>
> node1:~ hiemstra$ ssh node2 which hboot
> no hboot in "/bin /sbin /usr/bin /usr/sbin /usr/local/bin"
>
> node1:~ hiemstra$ ssh node2
> Welcome to Darwin!
> node2:~ hiemstra$ which hboot
> /usr/local/bin/hboot
>
> What on earth could be causing that?
>
> I've since reinstalled every node with a new build of lam-7.1.2b22 but
> had
> no luck in fixing this problem.
>
> Tony Arcieri
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
{+} Jeff Squyres
{+} jsquyres_at_[hidden]
{+} http://www.lam-mpi.org/
|
